ChatGPT now handles 2.5B queries daily

Today, we’re looking at ChatGPT’s 2.5 billion daily queries now rivaling search engines. Also in the mix: TikTok launches a real-world coding benchmark, the US moves on national AI rules and more.

In today’s post:

• ChatGPT hits 2.5B daily queries and what that really means

• TikTok’s SWE-Perf benchmark tests LLMs on real codebases

• The U.S. lays groundwork for national AI standards

What’s Trending Today

Together with TestPrite:

Meet the AI Testing Agent That’s Quietly Fixing Your Broken Features

AI is building your product.

But who’s checking its work?

Most vibe coding tools like Copilot or Cursor don’t crash.

They just quietly skip critical features auth flows, payment logic, dashboards that actually work.

TestSprite MCP changes that.

It’s the first AI testing agent that supervises your coding AI, catching invisible failures before they ship.

Here’s what it does fully automatic, no code, no prompts:

→ Reads your product intent (even writes a PRD if you don’t have one)
→ Builds a smart test plan
→ Finds what’s broken and missing
→ Loops with your AI agent until it’s fixed

In real-world benchmarks, feature accuracy jumped from ~42% to over 90%.

✅ No config
✅ No test writing
✅ Just working code

TestSprite MCP isn’t just a dev tool.

It’s the future of AI-native engineering and it already hit #1 on Product Hunt.

Try it now at TestSprite MCP while it’s still free.

Update

OpenAI’s AI assistant is now handling more queries than ever

OpenAI told Axios that ChatGPT now processes 2.5 billion prompts per day. That includes 330 million daily from U.S. users alone, a sign of just how embedded the tool has become in daily workflows.

Here’s everything you need to know:

• ChatGPT’s usage has more than doubled since December, when daily prompts were just over 1 billion.

• The new volume puts it within range of Google’s daily search numbers, estimated between 13.7 and 16.4 billion.

• Google doesn’t release official daily data, but recent reports suggest 5 trillion annual searches.

• The rapid rise of ChatGPT hints at a shift in how people seek information and solve problems.

• Unlike traditional search, ChatGPT offers conversational answers, coding help, and content generation.

• This surge also underscores how quickly generative AI is scaling in consumer and enterprise settings.

• With growing user demand, OpenAI is likely investing more into infrastructure and advanced models.

Here’s what I think about this:

We’re watching the early stages of a new digital behavior: searching less, prompting more. Whether this trend continues or levels off will depend on how well these tools keep up with real-world needs and how they earn trust as daily companions.

Policy

AI Policy Is Going National: Can Global Safety Keep Up?

The tone of international AI policy has changed dramatically in 2025. Countries are now prioritizing national competitiveness over collaborative safety, raising key questions about governance, regulation, and the future of global AI standards.

Here’s everything you need to know:

• The U.S. and U.K. have shifted from AI safety to national security and economic growth.

• France’s AI Action Summit signaled this pivot, with leaders calling for acceleration over caution.

• The U.S. repealed prior executive orders on AI safety, embracing deregulation and adoption.

• EU efforts to simplify AI regulation have triggered concerns from businesses about compliance gaps.

• State-level AI laws in the U.S. are being stalled or stripped down due to industry pressure.

• Despite this, global forums like the AI Impact Summit in New Delhi offer a chance to restore focus on trust, evaluation, and collaboration.

• Experts say nations must still align on four fronts: regulatory gaps, open-source standards, robust testing, and public-private coordination.

This isn’t the end of international AI cooperation but it’s a wake-up call. Speed without structure invites risk. If countries want to lead in AI, they also need to lead in setting smart, interoperable standards. Innovation and governance must evolve together or neither will hold.

Breakthrough

TikTok just dropped a benchmark that could reshape how we trust AI in code

TikTok researchers and collaborators have launched SWE-Perf the first benchmark built to test whether large language models can meaningfully optimize real-world code performance.

This isn’t about fixing bugs. It’s about tuning entire repositories.

• Most LLM benchmarks stop at syntax and correctness SWE-Perf goes deeper.

• The goal: measure how well models can improve performance at the repo level, not just in isolated functions.

• It draws from 100,000+ pull requests and includes expert-verified patches, Dockerized testbeds, and runtime validation.

• Two modes exist: “oracle” for scoped optimization and “realistic” for repo-wide tuning.

• Top models like Claude and GPT-4o scored under 2% performance gains far behind human-level tuning (10.85%).

• The standout? An agent-based setup called OpenHands, built on Claude-3.7, showing promise in realistic conditions.

• Key takeaway: LLMs handle low-level tweaks decently but struggle with big-picture performance patterns.

Here’s what I think:

This is a needed step forward. If we want AI to move from autocomplete to co-pilot, it has to prove it can improve systems, not just syntax. SWE-Perf gives us a clearer lens but it also shows we’re still early. Optimizing software at scale might be the last mile for AI engineering.

🚀 Founders & AI Builders, Listen up!

If you’ve built an AI tool, here’s an opportunity to gain serious visibility.

Nextool AI is a leading tools aggregator that offers:

• 500k+ page views and a rapidly growing audience.

• Exposure to developers, entrepreneurs, and tech enthusiasts actively searching for innovative tools.

• A spot in a curated list of cutting-edge AI tools, trusted by the community.

• Increased traffic, users, and brand recognition for your tool.

Take the next step to grow your tool’s reach and impact.

Reach 140,000+ READERS:

Expand your reach and boost your brand’s visibility!

Partner with Nextool AI to showcase your product or service to 140,000+ engaged subscribers, including entrepreneurs, tech enthusiasts, developers, and industry leaders.

Ready to make an impact? Visit our sponsorship website to explore sponsorship opportunities and learn more!